利用Gleeble-1500D试验机对新型Mg-Sm-Zn-Zr合金进行等温压缩实验,得到了该合金在350～450℃、0.001～1 s-1条件下的真应力-应变曲线,应用遗传算法优化的BP神经网络建立起合金的应力预测模型,并对所建预测模型和考虑应变的Ar-rhenius本构模型进行了对比,采用预测数据并应用Murthy失稳准则绘制出该合金的热加工图,最后结合微观组织分析所绘制热加工图的合理性.结果表明,GA-BP模型预测值和实验值间的相关性系数为o.999,平均相对误差为1.469％,较应变补偿本构模型预测精度更高;热加工图设计合理,有效确认温度400～450℃、应变速率0.001～0.03 s-1是最佳热加工范围,合金在该区域发生了动态再结晶.
The flow stress behavior of Mg-Sm-Zn-Zr alloy was studied by isothermal compression experiment on Gleeble-1500D thermal-mechanical test machine at deformation temperatures of 350-450 ℃ and strain rates of 0.001-1 s-1.The genetic algorithm BP neural network (GA-BP) was developed to predict the flow stress,and the comparative study on GA-BP model and strain compensated Arrhenius-type constitutive model was presented.Based on the prediction stress,the processing map was established under instability criteria of Murthy,finally the rationality of the designed processing map was verified by microstructure.The results showed that the correlation coefficient was 0.999 and the average relative error was 1.469％ for the GA-BP model,which indicated that the GA-BP model could be more accurate in predicting the flow stress than constitutive model considering the compensation of strain.The processing map was properly designed,and the map confirmed the temperatures of 400-450 ℃ and strain rates of 0.001-0.03 S-1 as the optimum process parameters.The dynamic recrystallization (DRX) occurred in the deformed samples under the above parameters.
参考文献
- 下载量()
- 访问量()
- 您的评分：
-
10%
-
20%
-
30%
-
40%
-
50%